If perhaps you're interested in the opinion of someone who knows more about
census geography/data than about TAZs specifically:
I think TAZs should nest to census tracts (which by definition will then
follow county lines). Tracts need more than one TAZ where there are
employment concentrations; more than one tract may constitute a TAZ when
you're in homogenous residential areas within a municipality.
I also think you should be worrying, or at least thinking, about the
sampling errors involved in small population TAZs. Block group data, even
with an average of 600 housing units, are too small to deliver decent data.
This is especially true for the journey to and place of work data, which
have additional non-response errors beyond the overall sampling and
non-response errors for the long form. I have never reported or relied on
census-based place of work counts unless there were at least 1000 workers
in the geographic area being reported.
My guess is that census will end up permitting data to be delivered on a
CTPP/UTPP for whatever geography you draw; disclosure will probably not be
a problem unless you start using the finely drawn race/ethnic groups which
will now be possible. But just because you can get the data doesn't mean
the numbers are worth anything.
One final word--at the risk of incurring the wrath of at least some of
you--I really don't see why it's so important to maintain complete
comparability from one set of TAZs to the next. The most important thing,
I would think, is to have TAZs which reflect the current land
use/population distribution/etc. to make good forecasts. If they need
changing for whatever reason, they should be changed.
Patricia C. (Patty) Becker 313/535-2077
APB Associates/SEMCC FAX 313/535-3556
17321 Telegraph #204 Home 248/355-2428
Detroit, MI 48219 pbecker(a)umich.edu
Show replies by date